Vol. 58: 311–321, 2010 AQUATIC Published online February 11 doi: 10.3354/ame01367 Aquat Microb Ecol

Genes for transport and metabolism of spermidine in pomeroyi DSS-3 and other marine

Xiaozhen Mou1,*, Shulei Sun2, Pratibha Rayapati2, Mary Ann Moran2

1Department of Biological Sciences, Kent State University, Kent, Ohio 44242, USA 2Department of Marine Sciences, University of Georgia, Athens, Georgia 30602, USA

ABSTRACT: Spermidine, putrescine, and other polyamines are sources of labile carbon and nitrogen in marine environments, yet a thorough analysis of the functional genes encoding their transport and metabolism by marine bacteria has not been conducted. To begin this endeavor, we first identified genes that mediate spermidine processing in the model marine bacterium and then surveyed their abundance in other cultured and uncultured marine bacteria. R. pomeroyi cells were grown on spermidine under continuous culture conditions. Microarray-based transcriptional profiling and reverse transcription-qPCR analysis were used to identify the operon responsible for spermidine transport. Homologs from 2 of 3 known pathways for bacterial polyamine degradation were also identified in the R. pomeroyi genome and shown to be upregulated by spermidine. In an analysis of genome sequences of 109 cultured marine bacteria, homologs to polyamine transport and degradation genes were found in 55% of surveyed genomes. Likewise, analysis of marine meta- genomic data indicated that up to 32% of surface ocean bacterioplankton contain homologs for trans- port or degradation of polyamines. The degradation pathway genes puuB (γ-glutamyl-putrescine oxi- dase) and spuC (putrescine aminotransferase), which are part of the spermidine degradation pathway in R. pomeroyi, emerged as suitable targets for molecular-based studies of polyamine pro- cessing by marine bacterial communities. The frequency of genes encoding transport and catabolism of spermidine and related polyamines suggests an important role for these compounds in carbon and nitrogen budgets of marine bacterioplankton.

KEY WORDS: Polyamine · Transcriptomic analysis · Microarray · Marine bacteria · Dissolved organic nitrogen

Resale or republication not permitted without written consent of the publisher

INTRODUCTION (Tyms 1989, Marian et al. 2000). As free constituents in seawater, however, they are only found at nM levels Spermidine and other polyamines are aliphatic (Jorgensen et al. 1993, Lee & Jorgensen 1995, Nishi- organic compounds with multiple amino groups. They bori et al. 2001, 2003), partly due to their active are synthesized by organisms across all 3 domains of turnover by marine bacteria (Höfle 1984, Lee & Jor- life, playing vital roles in diverse cellular processes gensen 1995). including nucleic acid and protein biosynthesis (Tabor Compared to extensive studies on the concentration & Tabor 1985, Higashibata et al. 2000, Kusano et al. and fate of other dissolved organic nitrogen com- 2007) and biosilica precipitation in diatom frustule for- pounds in seawater, such as dissolved free amino acids mation (Kroger et al. 2000, Sumper & Kroger 2004). In (DFAA), investigations of bacterially-mediated poly- the cytoplasm of bacteria and marine algae, intracellu- amine transformations have been rare (Höfle 1984, Lee lar soluble polyamine concentrations reach mM levels & Jorgensen 1995). Yet recent metagenomic and meta-

*Email: [email protected] © Inter-Research 2010 · www.int-res.com 312 Aquat Microb Ecol 58: 311–321, 2010

transcriptomic sequencing of marine microbial com- normalized to 3 mM carbon (i.e. 0.43 mM spermidine, munities have recovered homologs of genes predicted 0.75 mM putrescine, or 1 mM serine, Table 1). Cells to be involved in spermidine and putrescine transport were grown in 200 ml chemostats at 30°C, a dilution and metabolism (Venter et al. 2004, Poretsky et al. rate of 0.125 h–1, an airflow rate of 1 ml min–1, and a 2005, Mou et al. 2008), in some cases showing evi- stirring speed of 200 rpm. Chemostat cultures were dence of differential distribution with ocean depth maintained at a constant cell density of OD600 = 0.2 for (DeLong et al. 2006). Furthermore, a recent Sargasso at least 4 retention times prior to harvesting. To harvest Sea study identified spermidine transporter proteins the cells, the outflow pumps were set to 6 ml min–1, and from SAR11 as an abundant component of the surface 9 ml of cells were directly collected in chilled tubes seawater metaproteome (Sowell et al. 2009). This containing 1 ml stop solution (5% phenol, 95% sequence-based evidence suggests that polyamines ethanol, pH = 8). Chemostat cultures were established are important substrates for heterotrophic microorgan- in quadruplicate for all 3 compounds in order to pro- isms in the ocean, with implications for the cycling of vide 4 independent replicates of each treatment for both nitrogen and carbon. microarray or qPCR analysis. In the present study, we focused on spermidine use RNA extraction, purification, and amplification. by the marine Roseobacter member Ruegeria Cells harvested from the chemostat cultures were pomeroyi DSS-3. Up to now, systematic investigations immediately centrifuged at 5000 × g (10 min) at 4°C; of spermidine transport and catabolism have been the cell pellets were then frozen at –80°C or used restricted to a few model bacteria with clinical and immediately for RNA extraction. Total RNA extraction, medical implications, such as Escherichia coli K12 mRNA purification, and mRNA amplification to amino- (Shaibe et al. 1985) and Pseudomonas aeruginosa allyl labeled antisense RNA (aa-aRNA) were per- PAO1 (Lu et al. 2002, Dasu et al. 2006). However, formed following protocols described previously in the Roseobacter lineage (Bürgmann et al. 2007). are prevalent in marine surface waters, where Microarray hybridization and processing. The they account for up to 20% of the total bacterio- Ruegeria pomeroyi DSS-3 whole genome microarray plankton in coastal areas and 10% in open oceans was designed on a CombiMatrix Custom Array plat- (Giovannoni & Rappe 2000). The genome sequence of form (Bürgmann et al. 2007). Along with probes for R. pomeroyi contains a number of genes that might be quality control, each array contains 12 000 probes that involved in spermidine transport (Moran et al. 2004) target 4161 out of 4348 identified genes in the and metabolism. R. pomeroyi genome (mostly 2 probes per gene; Bürg- Transcriptional profiling using whole genome mann et al. 2007). The genes that were excluded from microarray analysis (Bürgmann et al. 2007) and qPCR the array either had close homologs in the genome or was used to identify key genes for spermidine process- did not meet probe design criteria with regard to ing in Ruegeria pomeroyi, and to provide insights into hybridization temperature. transporter specificity. A comprehensive bioinformatic The aa-aRNA was fluorescently labeled and survey of polyamine-related genes in marine bacterial hybridized to the Ruegeria pomeroyi microarray as genomes and metagenomes confirmed the numerical described previously (Bürgmann et al. 2007), except abundance of polyamine transport and degradation that a non-competitive hybridization scheme was used, genes among marine bacterioplankton, and identified 2 candidate genes for Table 1. Structure and C:N ratio of 5 polyamine compounds used in the present monitoring of polyamine transforma- study and the amino acid serine tions in situ. Compound Formula C:N Chemical Structure

MATERIALS AND METHODS Cadaverine C5H14N2 2.5 H2N NH2

Norspermidine C H N 2 H N NH Culture conditions. Ruegeria pome- 6 17 3 2 NH 2 royi DSS-3 cells were grown in a modi- Putrescine C4H12N2 2 H2N NH fied marine basal medium (MBM; Gon- 2 NH NH zalez et al. 2003) containing spermidine Spermidine C7H19N3 2.33 H2N 2 NH or putrescine as the sole carbon and NH Spermine C H N 2.5 H N 2 nitrogen source. Serine was also used 10 26 4 2 NH as a substrate to provide comparative O expression data for an amino acid. The Serine C3H7NO3 3 HO OH concentration of the 3 compounds was NH2 Mou et al.: Polyamine-related genes in marine bacteria 313

i.e. aa-aRNA was labeled only with a single dye (Alex- were designed using Geneious Pro software (Bio- aFluor dye 647; Invitrogen). After hybridization, the matters) and are listed in Table S1 available as supple- microarrays were scanned with an Axon GenPix 4000B mentary material at www.int-res.com/articles/suppl/ microarray scanner (Molecular Devices Corporation) a058p311_app.pdf. The designed annealing tempera- at 5 µm resolution. Images were acquired and ana- ture for each primer set ranged from 59 to 61°C. The lyzed using GenePix Pro 6.0 software (Molecular practical annealing temperature for all primer sets was Devices Corporation). The detection limits (DL) were chosen by performing a gradient PCR assay (annealing calculated based on reading of the 149 empty spots on temperature varied between 57 and 65°C) using the the array using the equation: DL = average sum of genomic DNA of Ruegeria pomeroyi as template, medians + 2×(SD). Spots with intensity below the DL and this converged at 60°C. Both aa-aRNA and non- and those identified as bad, empty, and not meeting amplified mRNA extracts that had only been treated quality assurance were excluded from further analysis. for rRNA removal (mRNA only kit; Invitrogen) were Background corrected expression data from each used as RT-qPCR templates. RNA samples were quan- array were globally normalized by trimmed means tified by spectrophotometer and were reverse tran-

(2% from each side) and log2 transformed prior to scribed to cDNA with random hexamer primers at a being imported into the Acuity 4.0 software (Molecular concentration of 0.3 µg µl–1 using iScript (Bio-Rad) Devices Corporation). Analysis datasets were created according to the manufacturer’s instructions. Triplicate using the conditions (signal-to-noise ratio > 3; circular- qPCR reactions were conducted in 25 µl volumes on an ity > 80; F635% < 2; B635 CV < 50) to exclude probes iCycler IQ multicolor Real-Time PCR detection System with features close to background, saturated, with bad (Bio-Rad). Each reaction contained 1 µl of cDNA tem- circularity, or with highly non-uniform intensities or plate or standard, forward and reverse primers at a background. final concentration of 10 pM each, and 1× IQ SYBR Gap statistics predicted that the optimal cluster size Green Supermix (Bio-Rad). Control reactions omitting for the array data was 3. Self Organizing Maps (SOM) template or reverse transcriptase were included for cluster analysis was then performed within Acuity each analysis. The qPCR program contained an initial using a 1×3 cluster matrix, the Euclidean squared denaturation step (95°C, 5 min) and 45 amplification similarity metric, and data centering (Fig. S1 available cycles, each consisting of a denaturation step (95°C, as supplementary material at www.int-res.com/ 45 s), annealing step (60°C, 90 s), and a final melting articles/suppl/a058p311_app.pdf). The 4 experimental curve analysis. Standards were obtained from a replicates were averaged to calculate the fold change dilution series of PCR amplicons of RNase-treated between the spermidine samples and serine controls. R. pomeroyi genomic DNA over 6 orders of magnitude t-test parameters and false discovery rates (FDR) were for each primer set and used for calculating fold calculated to determine the significance of observed changes between treatments. For transporter substrate differences. Upregulated genes were reported when preference analysis, an identical RT-qPCR procedure gene probes in spermidine samples showed more than was performed except that only the gene sets for the 2-fold increase in expression level than those in serine 6 polyamine-binding protein genes were used and controls, with t-test p values < 0.05, false discovery rate mRNA was obtained from both spermidine- and < 10%, and membership in SOM probe clusters that putrescine-grown chemostat cultures. showed increased expression in the spermidine rela- Phylogenetic analysis. The amino acid sequences of tive to the serine samples. Downregulated genes were the putative polyamine substrate-binding proteins in reported when genes showed an opposite but equiva- Ruegeria pomeroyi were aligned based on Blosum62 lent response. The remainder of the genes were desig- using ClustalW with the MEGA4 program (Kumar et nated as non-responding. Microarray data were al. 2004) and checked manually. The resulting distance deposited in the Gene Expression Omibus database matrix was used for generating a phylogenetic tree (www.ncbi.nlm.nih.gov) under accession number using the minimum evolutionary, neighbor-joining, GLP4067. and UPGMA algorithms with 1000 bootstraps using RT-qPCR. For microarray result verification and MEGA4. transporter preference analysis, 12 genes were ana- Polyamine growth survey. Ruegeria pomeroyi DSS- lyzed by reverse transcription (RT)-qPCR using mRNA 3 cells were grown in MBM containing the following obtained independently from that used for microarray single polyamine compounds as sole carbon and nitro- analysis. The genes included each of the polyamine- gen sources: spermidine, putrescine, cadaverine, binding components of 6 complete polyamine ABC norspermine, and spermine. Serine provided compara- transporter systems and another 2 genes randomly tive data for an amino acid. The concentration of each selected from each of the upregulated, downregulated, compound was normalized to 3 mM carbon (i.e. and non-responding genes. Primer sets for each gene 0.43 mM spermidine, 0.75 mM putrescine, 0.6 mM 314 Aquat Microb Ecol 58: 311–321, 2010

cadaverine, 0.5 mM norspermidine, 0.3 mM spermine, was adopted for the identification of homologs to R. and 1 mM serine; Table 1). Cells were grown in the pomeroyi SPO3465 (predicted as puuB) and SPO3473 dark at 30°C with shaking at 200 rpm. Biomass was (predicted as spuC) in the GOS dataset. The final cut- measured as the optical density of cells at a wave- off for all analyses converged on an E value of <10–30. length of 600 nm (OD600) at regular intervals until the The acetylornithine aminotransferase gene (argD) cells reached a stationary phase. Batch cultures were homolog in R. pomeroyi (SPO0962) has a related but established in triplicate. distinct function to spuC. The blast hits that had a Bioinformatic analysis. Polyamine transport systems higher bit score for argD than spuC in blastp analyses (gene designation: pot) were putatively identified from were removed from the final list of spuC homologs. sequenced marine bacterioplankton genomes in the Paired read sequences in GOS data were only counted Moore Microbial Genome sequencing database (data once. freeze date: 1 October 2008; https://moore.jcvi.org/ Homologs to polyamine synthesis genes in Ruegeria moore/) based on key word searches. A blastp assay pomeroyi were obtained by blasting known genes from was also performed using identified polyamine-bind- E. coli K12 (Blattner et al. 1997, http://ecoli.naist.jp/ ing proteins in the genome of Escherichia coli K12 GB6/search.jsp) and Pseudomonas aeruginosa PAO1 (potD and potF ), with an E value cutoff of <10–20. The (Stover et al. 2000, www. pseudomonas.com) to the R. candidates from the 2 procedures were combined and pomeroyi genome sequence using blastp through the then manually inspected for orthology to polyamine RoseoBase website (www.roseobase.org). Gene ortho- transporters and to ensure that each identified pot sys- logs were reported when the reciprocal best hit E tem contained consecutive genes that encoded at least value was <10–30. 1 copy of each of the 4 components, which is required for a polyamine transporter to function. Polyamine degradation genes do not require consecutive gene RESULTS systems to function. A reciprocal best hit methodology with an E value cutoff of <10–30 was used to identify orthologs to puuB and spuC in the sequenced marine Microarray analysis and quality control bacterioplankton genomes in the Moore genome data- base. Ruegeria pomeroyi cells were grown on spermidine Homologs to experimentally confirmed polyamine- or serine under steady-state conditions in a chemostat binding proteins (potD and potF ) were identified in with fixed cell growth rates, temperature, pH, and air the Global Ocean Survey (GOS) dataset using blastp flow. This effort minimized gene expression artifacts with an E value cutoff of <10–20. The 10 GOS hits at the due to factors other than substrate differences. Of the boundary of the cutoff were blasted back to the NCBI 4161 genes that were arrayed, significant increases in Refseq database (www.ncbi.nlm.nih.gov/RefSeq/) and mRNA levels were found for 92 genes in spermidine the Escherichia coli and Ruegeria pomeroyi genomes. treatments relative to serine treatments (i.e. ≥2-fold If 2 or more hits were not to the correct functional cat- higher normalized fluorescence; t-test, p < 0.05; egory, the E value cutoff was decreased by 5 orders of Table 2); these were designated as upregulated genes magnitude for a subsequent blast. The same procedure (Fig. 1). About one-third of the upregulated genes

Table 2. Ruegeria pomeroyi. Upregulated genes hypothesized to be involved in spermidine transport and degradation by R. pomeroyi. Complete lists of up- and downregulated genes are provided in Tables S2 & S3 available as supplementary material at www.int-res.com/articles/suppl/a058p311_app.pdf

Gene locus tag COG Gene name Annotation

SPO1300 COG0174 Glutamine synthetase family protein SPO1301 COG0518 Glutamine amidotransferase class I SPO1302 COG0174 Glutamine synthetase family protein SPO2659 COG0765 gltJ Glutamate/aspartate ABC transporter, permease protein SPO3465 Conserved hypothetical protein SPO3466 COG1177 potI Putrescine ABC transporter, permease protein SPO3467 COG1176 potH Putrescine ABC transporter, permease protein SPO3468 COG3842 potG Putrescine ABC transporter, ATP-binding protein SPO3469 COG0687 potF Putrescine ABC transporter, periplasmic substrate-binding protein SPO3471 COG0161 Aminotransferase class III SPOA0273 DNA-binding protein, putative Mou et al.: Polyamine-related genes in marine bacteria 315

6

5

4 3465 1323

p value) p 1300 1301 2659 3469 10 3

log 1302 3466 2 3468 3467 Fig. 1. Ruegeria pomeroyi. Transcriptional response of genes grown on spermidine 1 relative to serine shown on a volcano plot. Up- or downregulated genes (≥2 fold Significance (– expression change; p < 0.05) in spermidine 0 treatments relative to the serine treatments are illustrated as open or gray circles, with locus tag coding of the gray circles the same as in Tables 2, S2, & S3. Genes show- –4 –3 –2 –1 0 1 2 3 4 ing no significant transcriptional changes Gene expression fold change (log2 spermidine/serine) are illustrated as black circles

(29 genes) had sequences too divergent to allow anno- ATP-binding protein. The genes of the substrate- tation of even general function (Table S2 provided as binding proteins are the most divergent, while the supplementary material at www.int-res.com/articles/ genes of the ATP-binding proteins are the most con- suppl/a058p311_app.pdf). Most of the remainder served (Tam & Saier 1993, Saurin & Dassa 1994). represented functions involved in substrate transport In the Ruegeria pomeroyi genome, 6 complete sets of (12 genes), nitrogen metabolism (12), carbon metabo- 4-component pot systems have been predicted based lism (10), and protein regulation (10). As expected, the on sequence homology (Moran et al. 2004). Only 1 set, L-serine ammonia-lyase gene (sdaA), a gene for serine SPO3466–SPO3469, was upregulated by exogenous degradation, was downregulated in the spermidine spermidine in the microarray analysis (Table 2) and, treatment relative to serine (t-test, p < 0.05), as were other serine metabolism- 8 related genes (Table S3). RT-qPCR using original mRNA extracts and amino-allyl 7

labeled antisense RNA (aa-aRNA, the SPO3469 same form as the microarray template) 6 agreed well with each other and both agreed with the microarray data (Fig. 2). 5 SPO2860

4

Transport genes upregulated by SPO3606 3 SPO3040

spermidine SPO1606 SPO2701 SPO1830 SPO2007 2 SPOA0381 SPO2578

In bacteria, exogenous polyamines are SPO3473 thought to be mainly transported by ATP- SPO0611

1 SPO1323 binding cassette transport systems (ABC- type transporters; Tabor & Tabor 1985). Fold change by qPCR (spermidine/serine) 0 Each polyamine transport system (gene 0 0.5 1 1.5 2 2.5 3 designation: pot) typically consists of 4 in- Fold change by microarray (spermidine/serine) dispensable components that are encoded by contiguous genes, i.e. 1 periplasmic Fig. 2. Ruegeria pomeroyi. Validation of microarray data shown by correla- tions with RT-qPCR-based analysis of expression level changes relative to substrate- binding protein, 2 hydrophobic serine. Closed circles, solid trend line: mRNA; open squares, dashed trend integral membrane proteins (permeases), line: amino-allyl labeled antisense RNA. Gene locus tags are labeled and 1 hydrophilic peripheral membrane adjacent to corresponding data points (see Tables 2, S2 & S3) 316 Aquat Microb Ecol 58: 311–321, 2010

unexpectedly, this system has been annotated as a for chemostat-grown cells with spermidine or putrescine-specific transport system with gene desig- putrescine as the substrate. SPO3469 had 6.5-fold nations potFGHI. Comparing the amino acid sequence increased transcription during growth on spermidine, of the substrate-binding protein of the upregulated in accordance with the microarray data. SPO3473 and transporter system, SPO3469, to experimentally con- SPO1606 showed no upregulation in response to sper- firmed substrate-binding proteins in other bacteria, we midine, but had 3.2- and 4.5-fold increased transcrip- found high similarity to both putrescine and spermi- tion during growth on putrescine (Table 3). SPO2007, dine-binding proteins. SPO3469 was 50% identical in SPO2701, and SPOA0381, the putative polyamine- amino acid sequence to potF (experimentally shown to binding protein genes in 3 other pot systems in the R. be specific for putrescine binding in Escherichia coli), pomeroyi genome (Fig. 3), were not significantly 30% identical to potD (experimentally shown to bind upregulated by either spermidine or putrescine, and both spermidine and putresine in E. coli, but with a likely transport other polyamines or related com- preference for the former; Igarashi & Kashiwagi 1999), pounds. Batch culture assays indicated that at least 3 and 51% and 47% identical to spuD and spuE, the 2 additional polyamines (cadaverine, norspermidine, preferential spermidine-binding proteins in a 5-com- and spermine; Table 1) could be transported and ponent pot system (SpuDEFGH) of Pseudomonas metabolized by R. pomeroyi (Fig. S3 available as aeruginosa (Lu et al. 2002). supplementary material at www.int-res.com/articles/ Amino acid sequence alignments were used to suppl/a058p311_app.pdf). assess residue conservation and phylogenetic related- ness of the binding protein upregulated in response to spermidine (SPO3469), the other 5 putative pot system Catabolic genes upregulated by spermidine binding proteins in Ruegeria pomeroyi, and the experimentally confirmed spermidine- and putrescine- Three pathways have been hypothesized previously binding proteins in Escherichia coli and Pseudomonas to mediate bacterial spermidine degradation (Fig. 4). aeruginosa. SPO3469 and another putative pot binding protein gene from R. pomeroyi potF [NP415375, E. coli] (SPO1606) consistently clustered with potD 57 and potF from E. coli and spuD and spuE 80 spuD [NP248991, P. aeruginosa] from P. aeruginosa (Fig. 3). However, 100 SPO3469 is the only polyamine-binding spuE [NP248992, P. aeruginosa] protein in the genome that 70 R. pomeroyi SPO3469 [YP168665, R. pomeroyi] is conserved for all 3 amino acid residues 99 experimentally demonstrated to be the potD [NP415641, E. coli] most critical to spermidine binding by potD SPO1606 [YP166847, R. pomeroyi] in E. coli (Igarashi & Kashiwagi 1999; Fig. S2). SPO2007 [YP167241, R. pomeroyi]

SPO3473 [YP168669, R. pomeroyi]

Substrate preference of Ruegeria 99 SPO2701 [YP167911, R. pomeroyi] pomeroyi pot systems 61 SPOA0381 [YP165208, R. pomeroyi] The high sequence divergence of polyamine-binding protein genes among 0.1 the 6 pot systems in the Ruegeria pomeroyi Fig. 3. Ruegeria pomeroyi. Phylogenetic tree based on amino acid sequences genome (averaging only 19% sequence of polyamine-binding proteins in experimentally confirmed pot systems and identity; Fig. 3) suggests differing sub- in 6 putative pot systems in the genome of R. pomeroyi. The tree was con- strate preferences. This idea is consistent structed using minimum evolution with 1000 bootstrap replications, and with the high redundancy of pot trans- bootstrap values >50% are indicated at the branch nodes. Scale bar indi- cates the amount of genetic change in terms of the number of amino acid porters (6 sets) in the R. pomeroyi genome, substitutions per site. GenBank accession numbers and source organisms are as well as the apparent expression of only shown in brackets. Bold font indicates R. pomeroyi spermidine and 1 of these in response to growth on spermi- putrescine transporter genes identified in the present study. Black circles dine. To obtain additional information on and diamonds indicate binding proteins shown experimentally to transport putrescine and spermidine, respectively. Gray arrows indicate binding pro- transporter specificity, the relative tran- teins with 3 conserved amino acids hypothesized to be critical for spermidine scriptional response of the 6 polyamine- binding (Igarashi & Kashiwagi 1999). Other tree-building algorithms binding genes was compared by RT-qPCR (UPGMA, neighbor joining) resulted in nearly identical branching patterns Mou et al.: Polyamine-related genes in marine bacteria 317

Table 3. Ruegeria pomeroyi. Average fold changes in abundance of transcripts for polyamine-binding protein genes of R. pomeroyi based on RT-qPCR of chemostat-grown cells in polyamine treatments relative to serine treatments. Each fold-change result was an average of 9 sets of data (3 RT-qPCR technical replicates × 3 chemostat samples)

Samples SPO1606 SPO2007 SPO2701 SPO3469 SPO3473 SPOA0381

Putrescine mRNAa 3.2 0.9 1.8 1.4 4.5 1.1 Spermidine mRNAa 1.1 1.2 1.9 6.5 1.3 1.7 Spermidine aa-aRNAb 1.1 1.7 0.7 7.4 0.6 0.9 amRNA used as template; bamino-allyl-labeled antisense RNA used as template

Two of them channel spermidine through putrescine puuA homologs is consistent with the Pseudomonas degradation pathways, and these are further divided aeruginosa genome (spuB and spuI, both of which into the transamination (Lu et al. 2002) and γ-glutamy- have been shown experimentally to function in sper- lation routes (Kurihara et al. 2005). The third pathway midine degradation; Lu et al. 2002). SPO1301 is anno- involves oxidative cleavage of spermidine into 4- tated as a glutamine amidotransferase (GATase), aminobutyraldehyde and 1, 3-diaminopropane before which mediates the biosynthesis of a variety of organic further degradation to cellular intermediates (Dasu et nitrogen compounds including nucleotides. Putative al. 2006). All 3 routes produce intermediates for the GATases in Escherichia coli (designated puuD) and P. tricarboxylic acid (TCA) cycle and other cellular pro- aeruginosa (designated spuA) have been shown to be cesses. The microarray data suggested that 2 of these involved in spermidine degradation (Lu et al. 2002, pathways are operational in Ruegeria pomeroyi. All Kurihara et al. 2005), but both have low identities key genes of the transamination and γ-glutamylation to SPO1301 (~15% identical in each case). While routes have putative homologs in the genome of R. SPO1300–SPO1302 are located distantly from pomeroyi, and as detailed below, many were upregulated during growth on sper- midine (Table 2, Fig. 4). Acetyl-polyamine Arginine Upregulated genes SPO3465 and Ornithine speA SPO3471 are located on either side of the speG speC bltD aphA/B upregulated pot transport system SPO3466– Agmatine SPO3469 (Fig. 5). SPO3465 shares 30% speB identity with the γ-glutamyl-putrescine speD Putrescine Spermidine oxidase gene of Escherichia coli (puuB; puuA* spdH γ Kurihara et al. 2005) in the -glutamylation puuB* spuC* route of putrescine degradation (Fig. 3). 1,3-diaminopropane SPO3471 shares 60% identity with the 4-aminobutyraldehyde putrescine aminotransferase gene (spuC), which catalyzes the removal of 1 of the 2 kauB 3-amino puuC propanaldehyde amino groups from putrescine (Lu et al. Glutamate 2002; Fig. 4). The adjacent gene SPO3470 puuD 4-aminobutyrate kauB had a higher expression during growth on puuE* β-alanine spermidine compared to serine (although it (gabT) did not meet the significance criteria) and is annotated as a GntR-type transcrip- Succinate semialdehyde tional regulator. The 7 consecutive genes Malonic semialdehyde gabD SPO3465– SPO3471 together appear to form a cluster that functions in both sper- Citrate midine uptake and degradation (Fig. 5). Succinate Acetyl-CoA The upregulated genes SPO1300 and gltA SPO1302 are both similar to the γ-glu- Oxaloacetate tamyl-putrescine synthase gene puuA (both 35% identical), which catalyzes the Fig. 4. Hypothesized spermidine degradation pathways in bacteria. Gene γ names with normal font: no homolog in Ruegeria pomeroyi; bold font: first step of the -glutamylation pathway in homolog in R. pomeroyi; bold font and asterisk: homolog upregulated putrescine degradation (Kurihara et al. by spermidine in R. pomeroyi; arrow with no gene label: gene has not yet 2005; Fig. 4). This observed duplication of been identified (modified after Dasu et al. 2006, Chou et al. 2008) 318 Aquat Microb Ecol 58: 311–321, 2010

SPO3465–SPO3471 in the genome of Ruegeria pomeroyi (Fig. 5), orthologs have been found in a sin- gle cluster in other genomes (Lu et al. 2002, Kurihara et al. 2005). Upregulation of SPO2659, a putative gluta- mate transport gene (gltJ ), might indicate a higher demand for glutamate in response to increased activity of the γ-glutamylation degradation route. Increased expression of many homologs of genes with demonstrated function in putrescine degradation indicates that spermidine is converted to putrescine in Ruegeria pomeroyi (Table 2, Fig. 4), as has been hypothesized in a Pseudomonas strain (Padmanabhan & Kim 1965). We found no homologs for the third path- SPO1609 ight gray indicates permease genes, and way for spermidine degradation involving direct Polyamine transporter

SPO3473 SPO3474 SPO3475 oxidative cleavage that has been demonstrated in P. aeruginosa (Dasu et al. 2006; Fig. 4). Overall, the

genome. Raised boxes indicate predicted operons. microarray and qPCR data indicate that R. pomeroyi transports exogenous spermidine into the cell by a

genome (SPO3465 through SPO3475; SPO1300 through dedicated ABC-type transporter, degrades it to

SPO3472 putrescine, and then ultimately to the TCA cycle inter-

R. pomeroyi mediate succinate. Ammonia, alanine, and glutamate are also generated along the degradation pathway

R. pomeroyi (Fig. 4), satisfying both carbon and nitrogen demands of the cell. Spermidine-putrescine transporter amino- Putrescine transferase

SPO1606 SPO1607 SPO1608 Polyamine genes in marine bacterial genomes and metagenomes

SPO3470 SPO3471 A bioinformatic survey of 109 marine heterotrophic regulator

Transcriptional bacterioplankton genomes identified a total of 196 complete polyamine transport systems in 74 genomes from many major bacterioplankton groups (Table 4). SPO3469 Complete transporter gene sets are found in all Acti-

SPO1302; and SPO1606 through SPO1609) nobacteria, Rhizobiales, Roseobacter, SAR11, Oceano- spirillales, and Vibrionales genomes (2, 4, 2, 2, and 10

synthase genomes, respectively). Groups with few or no trans- SPO1302 porters include Bacteriodetes and Planctomyces (15 Putative glutamyl SPO3468 and 2 genomes, respectively). Multiple pot systems are often found in the same genome. For example, roseo- bacters average >4 pot systems per sequenced ge- nome. Some bacterioplankton, such as the 2 SAR11 representatives, are equipped with only a single pot Polyamine transporter system. Because polyamine-binding proteins may not cluster according to substrate specificities in phyloge- puuA puuD puuA potD potC potB potA

Organization of the spermidine- and putrescine-related gene clusters in the netic analyses (Fig. 3), the substrates of these marine synthase SPO1300 SPO1301 bacterioplankton pot transporters could not be identi- Putative glutamyl SPO3466 SPO3467 fied further. pot systems were also found with high frequency in the GOS metagenomic dataset, based on blastp queries using experimentally-confirmed polyamine-

puuB potI potH potG potFbinding protein genes spuC (Venter et al. 2004, Rusch et al. SPO3465 oxidase Ruegeria pomeroyi. 2007; Table 5). Ratios of putative pot binding proteins to universal single-copy genes (Howard et al. 2008) -glutamyl-putrescine γ Within the clusters that contain genes to transportWithin polyamines, white color coding indicates polyamine-binding protein genes, l protein genes. These genes occur in 3 separate clusters the dark gray indicates ATP-binding Fig. 5. indicate that as many as 32% of surface marine bacte- Mou et al.: Polyamine-related genes in marine bacteria 319

Table 4. Frequency of genes for polyamine transport (pot) and Table 4 (continued) degradation (puuB and spuC) in 109 sequenced marine bac- terioplankton genomes. Table entries show the number of multi-gene systems (pot) or single genes (puuB and spuC) per Bacterioplankton taxa pot puuB spuC genome. Only taxa with positive results are listed. The nota- tion in parenthesis shows the number of positive genomes/ Vibrio fischeri MJ11 2 total number of genomes surveyed for each group. Vibrio shilonii AK1 4 1 1 Vibrio sp. MED222 2 1 Vibrio splendidus 12B01 2 1 Bacterioplankton taxa pot puuB spuC Vibrionales bacterium SWAT-3 2 1 Other gamma (3/6) Alphaproteobacteria Congregibacter litoralis KT71 1 1 Roseobacter (23/23) Marine gamma HTCC2080 1 1 Alpha proteobacterium HTCC2255 3 1 1 Reinekea sp. MED297 1 1 Loktanella vestfoldensis SKA53 1 1 Actinobacteria (2/2) Oceanicola batsensis HTCC2597 3 1 1 Janibacter sp. HTCC2649 1 1 Oceanicola granulosus HTCC2516 2 1 Marine actinobacterium PHSC20C1 2 1 Oceanibulbus indolifex HEL-45 3 1 1 Firmicutes (3/4) Octadecabacter antarcticus 307 4 1 Bacillus sp. B14905 1 Phaeobacter gallaeciensis 2.10 3 1 1 Bacillus sp. NRRL B-14911 1 Phaeobacter gallaeciensis BS107 2 1 1 Carnobacterium sp. AT7 1 bacterium HTCC2150 2 1 1 Bacteroidetes Rhodobacterales bacterium HTCC2654 2 1 1 Flavobacteria (1/12) Roseobacter litoralis Och 149 6 1 1 Psychroflexus torquis ATCC 700755 1 Roseobacter sp. AzwK-3b 3 1 1 Other bacteroidetes (1/3) Roseobacter sp. CCS2 2 1 1 Microscilla marina ATCC 23134 1 Roseobacter sp. MED193 2 1 1 Cyanobacteria (2/10) Roseobacter sp. SK209-2-6 4 1 1 Lyngbya aestuarii CCY9616 2 Roseovarius nubinhibens ISM 7 1 1 Prochlorococcus marinus MIT 9211 1 Roseovarius sp. 217 8 1 1 Planctomyces (0/2) Roseovarius sp. HTCC2601 3 1 1 Other (3/10) Roseovarius sp. TM1035 5 1 Caminibacter mediatlanticus TB-2 1 Ruegeria pomeroyi DSS-3 6 1 1 Marinitoga piezophila KA3 1 Sagittula stellata E-37 6 1 1 Plesiocystis pacifica SIR-1 1 Sulfitobacter sp. EE-36 2 1 1 Sulfitobacter sp. NAS-14.1 2 1 1 Other Rhodobacterales (2/3) Stappia aggregata IAM 12614 5 1 1 rioplankton cells could contain a pot system. This num- Stappia alexandrii DFL-11 2 1 1 ber is an overestimate of per-cell frequency if multiple SAR11 (2/2) Pelagibacter ubique HTCC1002 1 1 pot systems are present in some bacterioplankton Pelagibacter sp. HTCC7211 1 1 genomes, as is the case for cultured roseobacters Rhizobiales (4/4) (Table 4), or if 1 pot system contains >1 polyamine- Aurantimonas sp. SI85-9A1 1 1 binding protein, as is the case for Pseudomonas aerug- Fulvimarina pelagi HTCC2506 1 1 Hoeflea phototrophica DFL-43 2 1 1 inosa (Lu et al. 2002). This frequency is lower than that Nitrobacter sp. Nb-311A 1 found for pot systems in cultured marine bacterio- Other alpha (1/7) plankton (68%; Table 4). Alpha proteobacterium BAL199 4 1 1 (6/9) Alteromonas macleodii Deep ecotype 1 In situ probes for polyamine degradation Marinobacter algicola DG893 2 Marinobacter sp. ELB17 4 1 Homologs to Ruegeria pomeroyi spermidine degrada- Moritella sp. PE36 1 tion genes SPO3465 ( ) and SPO3473 ( ) (pu- Pseudoalteromonas tunicata D2 1 1 puuB spuC benthica KT99 1 1 trescine transamination and γ-glutamylation pathways, Oceanospirillales (3/3) respectively) were found with high frequency in se- Marinomonas sp. MED121 3 1 1 quenced marine bacterioplankton genomes and the sp. RED65 2 1 1 Oceanobacter GOS metagenomic dataset (Tables 4 & 5). Of the 109 Oceanospirillum sp. MED92 2 1 1 Vibrionales (10/10) genomes examined, 47 puuB and 36 spuC orthologs Photobacterium profundum 3TCK 2 1 were identified (reciprocal best hits in blastp analyses, Photobacterium sp. SKA34 3 1 Table 4). These putative orthologs are distributed widely Vibrio alginolyticus 12G01 4 1 among major marine bacterial taxa, and are particularly Vibrio angustum S14 4 1 Vibrio campbellii AND4 2 prevalent in the Roseobacter lineage. The 2 representa- tive bacteria in the SAR11 clade each had a single ho- 320 Aquat Microb Ecol 58: 311–321, 2010

Table 5. Frequency of genes for polyamine transport (substrate-binding proteins potD and potF) and degradation (puuB and spuC) in the Global Ocean Sampling (GOS) metagenomic data, expressed as homolog number (and % of cells ± SD), assuming no more than one gene copy per cell

GOS sitesa Genomes sampledb potD/F puuB spuC

Coastal (n = 19) 1222 409 (32 ± 12%) 108 (10 ± 7%) 189 (17 ± 11%) Open ocean (n = 13) 1334 500 (32 ± 11%) 183 (11 ± 7%) 340 (24 ± 16%) Hypersaline (n = 1) 298 44 (15%) 17 (6%) 40 (13%) Estuary and Other (n = 10) 962 206 (21 ± 14%) 39 (5 ± 7%) 66 (10 ± 11%) aGOS data also include 6 pilot Sargasso Sea datasets (Sites 2 to 7) bGenome equivalents sequenced at each GOS site are taken from Howard et al. (2008) and are based on numbers of homologs of the single-copy gene recA

molog of these genes (Table 4). Assuming a single copy The substrate specificity of polyamine transporters in per cell, puuB and spuC homologs are present in 10% marine bacterial genomes and metagenomes is hard to and 17% of bacterioplankton cells in coastal metage- decipher at present. Whole-genome microarray and nomic libraries, and 11% and 24% in open ocean li- RT-qPCR data indicate that Ruegeria pomeroyi trans- braries (Table 5; frequencies are not significantly differ- porter systems may have narrow substrate specifici- ent between coastal and open ocean GOS sites; t-test, p ties. This is further supported by the low identity > 0.05). For most marine bacterial genomes, puuB and (~19%) of amino acid sequences among predicted spuC co-occur with pot system genes for transport of polyamine-binding proteins (Fig. 3). The clustering exogenous polyamines into the cell (Table 4). pattern of binding proteins cautions against the use of sequence identity alone as an indicator of substrate specificity among the R. pomeroyi, Escherichia coli, DISCUSSION and Pseudomonas aeruginosa binding proteins ana- lyzed here (Fig. 3). Because of their high sequence Intracellular polyamines are found in virtually all living divergence, polyamine-binding proteins are not good organisms at cellular concentrations in the micromolar targets for the design of general qPCR primer sets for range. While this source of organic carbon and nitrogen environmental studies; even orthologs in genomes of should be valuable to marine heterotrophic bacteria, 2 closely related strains for which 16S rRNA genes are whose growth in seawater is often limited by carbon or ni- >96% identical (e.g. R. pomeroyi DSS-3 and Ruegeria trogen, bacterially-mediated polyamine transformation sp. TM1040) presented a significant challenge for has been studied in only a few cases. Results generated primer design. Our analyses suggest instead that the from the early studies are somewhat contradictory in well-conserved degradation genes puuB and spuC are terms of the fate of the carbon in exogenous polyamines. more robust targets for in situ probing of polyamine In one study, exogenous 14C-putrescine was preferen- processing by natural bacterial communities. tially respired to CO2 (~85%) rather than incorporated in Studies of ecologically-relevant model bacteria that biomass (~6%) by assemblages of marine bacterioplank- are designed and interpreted in the context of metage- ton (Höfle 1984). In another, the same polyamine ap- nomic data from natural bacterial communities can sig- peared to be largely incorporated into cell biomass by nificantly enhance our understanding of biogeochemi- bacterial assemblages in a coastal salt pond (40 to 75% in cal processes. Despite recognition of the importance of an oxic zone; 60 to 100% in an anoxic zone; Lee & Jor- spermidine and other polyamines in seawater over gensen 1995). In Ruegeria pomeroyi, spermidine is de- 2 decades ago, limited information had been gener- graded to intermediates that feed into the tricarboxylic ated on the biological processing of these compounds. acid cycle, from which both energy generation and This study provides new details on the function and biosynthesis are possible, and growth rates and biomass distribution of genes for transport and metabolism of yields are similar to those on the amino acid serine. The spermidine in a model marine bacterium. Moreover, it high concentrations used in the present study (up to 1 shows a high frequency and wide taxonomic distribu- mM) confirm that polyamines are not toxic to R. tion of homologs to polyamine transport genes (pot) pomeroyi, as was shown by Höfle (1984) for a marine bac- and degradation genes (puuB and spuC) in marine terial assemblage. Whether the fate of polyamine-derived bacterioplankton communities (Tables 4 & 5), arguing carbon and nitrogen (i.e. incorporation versus regenera- for increased attention to the role of polyamines in the tion) is affected by the concentration of polyamines and marine carbon and nitrogen cycles. Functional charac- the supply of other labile DOM is yet to be determined. terization of these and other ecologically relevant Mou et al.: Polyamine-related genes in marine bacteria 321

bacterial genes is a critical task for postgenomic envi- Advances in polyamine research in 2007. J Plant Res 120: ronmental microbiology. 345–350 Lee C, Jorgensen NOG (1995) Seasonal cycling of putrescine Acknowledgements. We thank J. Henriksen and W. Ye for and amino acids in relation to biological production in a assistance with the chemostats; R. Poretsky, S. Gifford, and stratified coastal salt pond. Biogeochemistry 29:131–157 J. Rinta-Kanto for helpful discussions on microarray data Lu CD, Itoh Y, Nakada Y, Jiang Y (2002) Functional analysis and analysis; C. English for graphic support; and G. Burnham and regulation of the divergent spuABCDEFGH-spuI operons C. Sloan for technical assistance. This research was supported for polyamine uptake and utilization in Pseudomonas aerug- by a National Science Foundation Grant to the Sapelo Island inosa PAO1. J Bacteriol 184: 3765–3773 Microbial Observatory (MCB0702125) and the Gordon and Marian FD, Garcia-Jimenez P, Robaina RR (2000) Polyamines Betty Moore Foundation. in marine macroalgae: levels of putrescine, spermidine and spermine in the thalli and changes in their concentra- LITERATURE CITED tion during glycerol-induced cell growth in vitro. Physiol Plant 110:530–534 Blattner FR, Plunkett G, Bloch CA, Perna NT and others Moran MA, Buchan A, Gonzalez JM, Heidelberg JF and oth- (1997) The complete genome sequence of Escherichia coli ers (2004) Genome sequence of Silicibacter pomeroyi K-12. Science 277:1453–1474 reveals adaptations to the marine environment. Nature Bürgmann H, Howard EC, Ye W, Sun F, Sun S, Napierala S, 432:910–913 Moran MA (2007) Transcriptional response of Silicibacter Mou X, Sun S, Edwards RA, Hodson RE, Moran MA (2008) pomeroyi DSS-3 to dimethylsulfoniopropionate (DMSP). Bacterial carbon processing by generalist in the Environ Microbiol 9:2742–2755 coastal ocean. Nature 451:708–711 Chou HT, Kwon DH, Hegazy M, Lu CD (2008) Transcriptome Nishibori N, Yuasa A, Sakai M, Fujihara S, Nishio S (2001) analysis of agmatine and putrescine catabolism in Pseudo- Free polyamine concentrations in coastal seawater during monas aeruginosa PAO1. J Bacteriol 190: 1966–1975 phytoplankton bloom. Fish Sci 67:79–83 Dasu VV, Nakada Y, Ohnishi-Kameyama M, Kimura K, Itoh Y Nishibori N, Matuyama Y, Uchida T, Moriyama T, Ogita Y, (2006) Characterization and a role of Pseudomonas aerug- Oda M, Hirota H (2003) Spatial and temporal variations in inosa spermidine dehydrogenase in polyamine catabo- free polyamine distributions in Uranouchi Inlet, Japan. lism. Microbiology 152:2265–2272 Mar Chem 82:307–314 DeLong EF, Preston CM, Mincer T, Rich V and others (2006) Padmanabhan R, Kim K (1965) Oxidation of spermidine by a Community genomics among stratified microbial assem- Pseudomonas sp. Biochem Biophys Res Commun 19:1–5 blages in the ocean’s interior. Science 311:496–503 Poretsky RS, Bano N, Buchan A, LeCleir G and others (2005) Giovannoni SJ, Rappe M (2000) Evolution, diversity, and mol- Analysis of microbial gene transcripts in environmental ecular ecology of . In Kirchman DL samples. Appl Environ Microbiol 71:4121–4126 (ed) Microbial ecology of the oceans. Wiley-Liss, Hobo- Rusch DB, Halpern AL, Sutton G, Heidelberg KB, Williamson ken, NJ, p 47–84 S (2007) The Sorcerer II global ocean sampling expedition: Gonzalez JM, Covert JS, Whitman WB, Henriksen JR and Northwest Atlantic through Eastern Tropical Pacific. PLoS others (2003) Silicibacter pomeroyi sp. nov. and Roseovar- Biol 5:e77 ius nubinhibens sp. nov., dimethylsulfoniopropionate- Saurin W, Dassa E (1994) Sequence relationships between demethylating bacteria from marine environments. Int J integral inner membrane proteins of binding protein- Syst Evol Microbiol 53:1261–1269 dependent transport systems: evolution by recurrent gene Higashibata H, Fujiwara S, Ezaki S, Takagi M, Fukui K, duplications. Protein Sci 3:325–344 Imanaka T (2000) Effect of polyamines on histone-induced Shaibe E, Metzer E, Halpern YS (1985) Metabolic pathway for DNA compaction of hyperthermophilic archaea. J Biosci the utilization of L-arginine, L-ornithine, agmatine, and Bioeng 89:103–106 putrescine as nitrogen sources in Escherichia coli K-12. J Höfle MG (1984) Degradation of putrescine and cadaverine in Bacteriol 163:933–937 seawater cultures by marine bacteria. Appl Environ Sowell SM, Wilhelm LJ, Norbeck AD, Lipton MS and others Microbiol 47:843–849 (2009) Transport functions dominate the SAR11 metapro- Howard EC, Sun S, Biers EJ, Moran MA (2008) Abundant and teome at low-nutrient extremes in the Sargasso Sea. ISME diverse bacteria involved in DMSP degradation in marine J 3:93–105 surface waters. Environ Microbiol 10:2397–2410 Stover CK, Pham XQ, Erwin AL, Mizoguchi SD and others (2000) Igarashi K, Kashiwagi K (1999) Polyamine transport in bacte- Complete genome sequence of Pseudomonas aeruginosa ria and yeast. Biochem J 344:633–642 PAO1, an opportunistic pathogen. Nature 406: 959–964 Jorgensen NOG, Kroer N, Coffin RB, Yang XH, Lee C (1993) Sumper M, Kroger N (2004) Silica formation in diatoms: the Dissolved free amino acids, combined amino acids and function of long-chain polyamines and silaffins. J Mater DNA as sources of carbon and nitrogen to marine bacteria. Chem 14:2059–2065 Mar Ecol Prog Ser 98:135–148 Tabor CW, Tabor H (1985) Polyamines in microorganisms. Kroger N, Deutzmann R, Bergsdorf C, Sumper M (2000) Microbiol Rev 49:81–99 Species-specific polyamines from diatoms control silica Tam R, Saier MH Jr (1993) Structural, functional, and evolu- morphology. Proc Natl Acad Sci USA 97:14133–14138 tionary relationships among extracellular solute-binding Kumar S, Tamura K, Nei M (2004) MEGA3: integrated soft- receptors of bacteria. Microbiol Mol Biol Rev 57:320–346 ware for molecular evolutionary genetics analysis and Tyms AS (1989) Polyamines and the growth of bacteria and sequence alignment. Brief Bioinform 5:150–163 viruses. In: Bachrach U, and Heimer YM (ed) The phys- Kurihara S, Oda S, Kato K, Kim HG, Koyanagi T, Kumagai H, iology of polyamines, Vol II. CRS Press, Boca Raton, FL, Suzuki H (2005) A novel putrescine utilization pathway p 3–33 involves gamma-glutamylated intermediates of Escheri- Venter JC, Remington K, Heidelberg JF, Halpern AL and oth- chia coli K-12. J Biol Chem 280:4602–4608 ers (2004) Environmental genome shotgun sequencing of Kusano T, Yamaguchi K, Berberich T, Takahashi Y (2007) the Sargasso Sea. Science 304:66–74

Editorial responsibility: Jed Fuhrman, Submitted: June 29, 2009; Accepted: August 19, 2009 Los Angeles, California, USA Proofs received from author(s): January 18, 2010